fueled by methane: deep-sea sponges from asphalt seeps...
TRANSCRIPT
The ISME Journal (2019) 13:1209–1225https://doi.org/10.1038/s41396-019-0346-7
ARTICLE
Fueled by methane: deep-sea sponges from asphalt seeps gain theirnutrition from methane-oxidizing symbionts
Maxim Rubin-Blum 1,2● Chakkiath Paul Antony1 ● Lizbeth Sayavedra1,3 ● Clara Martínez-Pérez 1
● Daniel Birgel4 ●
Jörn Peckmann 4● Yu-Chen Wu5
● Paco Cardenas 6● Ian MacDonald7
● Yann Marcon 8● Heiko Sahling9
●
Ute Hentschel5 ● Nicole Dubilier1,9
Received: 11 June 2018 / Revised: 16 December 2018 / Accepted: 20 December 2018 / Published online: 15 January 2019© International Society for Microbial Ecology 2019. This article is published with open access
AbstractSponges host a remarkable diversity of microbial symbionts, however, the benefit their microbes provide is rarelyunderstood. Here, we describe two new sponge species from deep-sea asphalt seeps and show that they live in a nutritionalsymbiosis with methane-oxidizing (MOX) bacteria. Metagenomics and imaging analyses revealed unusually high amountsof MOX symbionts in hosts from a group previously assumed to have low microbial abundances. These symbionts belongedto the Marine Methylotrophic Group 2 clade. They are host-specific and likely vertically transmitted, based on their presencein sponge embryos and streamlined genomes, which lacked genes typical of related free-living MOX. Moreover, genesknown to play a role in host–symbiont interactions, such as those that encode eukaryote-like proteins, were abundant andexpressed. Methane assimilation by the symbionts was one of the most highly expressed metabolic pathways in the sponges.Molecular and stable carbon isotope patterns of lipids confirmed that methane-derived carbon was incorporated into thehosts. Our results revealed that two species of sponges, although distantly related, independently established highly specific,nutritional symbioses with two closely related methanotrophs. This convergence in symbiont acquisition underscores thestrong selective advantage for these sponges in harboring MOX bacteria in the food-limited deep sea.
Introduction
Symbioses with microorganisms have played a central rolein shaping the ecology and evolution of marine animals [1].Sponges are one of the oldest animal phyla and may havelived in symbiosis with microbial partners for hundreds ofmillion years [2–4]. Most sponge species belong to one oftwo general categories: High microbial abundance (HMA)
Deceased: Heiko Sahling, 23 April 2018.
* Maxim [email protected]
* Nicole [email protected]
1 Max Planck Institute for Marine Microbiology, Celsiusstrasse 1,28359 Bremen, Germany
2 Israel Limnology and Oceanography Research, Tel Shikmona,3108000 Haifa, Israel
3 Quadram Institute Bioscience, Norwich Research Park,Norwich, UK
4 Institute for Geology, Center for Earth System Research andSustainability, University of Hamburg, 20146 Hamburg, Germany
5 GEOMAR Helmholtz Centre for Ocean Research, RD3 MarineMicrobiology and Christian-Albrechts University of Kiel,Düsternbrooker Weg 20, D-24105 Kiel, Germany
6 Department of Medicinal Chemistry, Pharmacognosy, BioMedicalCentre, Uppsala University, Husargatan 3, 751 23 Uppsala, Sweden
7 Florida State University, POB 3064326, Tallahassee, FL 32306,USA
8 Wegener Institute Helmholtz Centre for Polar and MarineResearch, HGF-MPG Group for Deep Sea Ecology andTechnology, Am Handelshafen 12, 27570Bremerhaven, Germany
9 MARUM, Center for Marine Environmental Sciences, Universityof Bremen, 28359 Bremen, Germany
Supplementary information The online version of this article (https://doi.org/10.1038/s41396-019-0346-7) contains supplementarymaterial, which is available to authorized users.
1234
5678
90();,:
1234567890();,:
sponges such as those from the orders Agelasida and Ver-ongida harbor dense microbial consortia with high phylo-genetic diversity, while low microbial abundance (LMA)sponges such as those from the order Poecilosclerida haveseveral orders of magnitude fewer symbionts with lowphylogenetic diversity [5]. In both HMA and LMA spon-ges, the microbial symbionts are hypothesized to increasetheir host’s fitness, for example by recycling nutrients andproducing secondary metabolites that can deter predators [4,6–8]. To date, the functional diversity of microorganismsassociated with sponges has been studied primarily in hostsfrom shallow-water habitats, using metagenomics [9–15],proteomics [16], and transcriptomics [17–21]. However, theremarkable diversity of the microbiota in most shallow-water sponges [3] has made it highly challenging tounderstand their functional and ecological role. Even less isknown about the metabolism, ecology, and evolutionaryhistory of the microorganisms that live in symbiosis withdeep-sea sponges.
In the deep sea, light is insufficient to sustain photo-synthetic primary production, the input of particulateorganic carbon from the surface is low, and nutrition isoften limited [22]. While shallow-water sponges primarilygain their nutrition by filter-feeding on planktonicmicroorganisms and organic matter [4], some spongeshave adapted to the deep sea, where low amounts ofparticulate organic carbon make filter-feeding energeti-cally unfavorable, by becoming carnivores [23–25].Sponges from cold seeps and hydrothermal vents in thedeep sea may have evolved an alternative nutritionalstrategy. In these environments, abundant chemosyntheticprimary production is fueled by reduced energy sources.Analyses of microbial communities based on metage-nomics and 16S rRNA gene amplicon sequencing, as wellas carbon stable isotope values, indicated that spongesfrom seeps are associated with chemosynthetic andhydrocarbon-degrading bacteria [25–30]. Thus, in addi-tion to filter-feeding or carnivory, these sponges couldgain a considerable proportion of their nutrition fromchemosynthetic symbionts.
In contrast to the well-studied chemosynthetic symbiosesof more highly evolved invertebrate groups such as bivalvesand annelids [31], little is known about these associations insponges. To gain a better understanding of chemosyntheticsymbioses in deep-sea sponges, we used metagenomics,metatranscriptomics, fluorescence, and electron micro-scopy, as well as lipid (fatty acids and sterols) and stableisotope analyses to study two sponge species from hydro-carbon seeps at Campeche Knolls at 2900–3100 m depth inthe southern Gulf of Mexico. These sites are characterizedby prolific asphalt flows, oil seepage, gas hydrates, and gasventing [32]. Campeche sponges were previously shown tohost hydrocarbon-degrading Cycloclasticus bacteria, but
these make up only about 5% of their bacterial community[33]. The majority of the sponge microbiota (50–70%) wasdominated by methane-oxidizing (MOX) bacteria(the abbreviation MOX is also used for methane oxidizer(s)in the following), [33]. In this study, we provide in-depthinsights into the symbiosis between sponges from theCampeche seeps and their MOX bacteria, with the goal ofbetter understanding the evolutionary history and physiol-ogy of the MOX symbionts, revealing the mechanismsthat might determine the specificity of the sponge-MOXassociation, comparing the genomic potential of the sym-biotic MOX with that of free-living MOX, and tracingthe incorporation of methane-derived carbon into spongebiomass.
Materials and methods
Sample collection
Sponges were collected with the remotely-operated vehi-cle MARUM-QUEST during the RV Meteor M114-2cruise to the Campeche Knolls in March 2015. Thesponges were collected by placing asphalt pieces onwhich they grew in an insulated polypropylene “bio-box”to protect against temperature changes during the ascen-sion of the ROV to warm surface waters (ascension timefrom seafloor to onboard ship ~2 h). We sampled twoencrusting sponge individuals, one from Chapopote Knoll(21°54′ N; 93°26′ W, 2925 m water depth) and one fromMictlan Knoll (22°1′ N; 93°14′ W, 3106 m water depth),which are ~25 km apart from each other (Fig. 1). A thirdsponge individual with a branching morphology wascollected from the same site at Chapopopote as theencrusting sponge. The sponges appeared healthy andintact before collection, with no evidence of tissuedamage. A detailed description of the collection sites isavailable elsewhere [32]. Sponge distributions at Chapo-pote were estimated using mosaic mapping (see Supple-mentary Methods 4).
After ROV recovery, the asphalt pieces with the spongeswere kept on board in the seawater from the “bio-box” at4 °C. For metagenomic and metatranscriptomic analyses,subsamples from each sponge individual were fixed onboard ~1 h after ROV recovery in RNAlater® (Sigma,Steinheim, Germany) according to the manufacturer’sinstructions and stored at −80 °C. Subsamples for micro-scopy were fixed in 2% paraformaldehyde in 1x phosphate-buffered saline (PBS) for at most 12 h at 4 °C, rinsedthree times in 1x PBS, and stored at 4 °C in 0.5x PBS/50%ethanol. Subsamples for transmission electron microscopywere fixed in PHEM buffer (PIPES, HEPES, EGTAand MgCl2, see ref. [34]). Samples for lipid biomarker
1210 M. Rubin-Blum et al.
analysis were flash-frozen in liquid nitrogen and storedat −80 °C.
Fluorescence in situ hybridization (FISH)
The probe used in this study, MTC849 probe (5′-CGTTAGCTCCACCACTAAG-3′), was designed with ARB[35] to target the 16S rRNA gene sequences of the symbioticMOX of both Campeche sponge species. This probe is amodification of MTC850 probe, designed to target MarineMethylotrophic Group (MMG) 2 MOX [36]. Apart from thesymbiotic and MMG2 MOX, the MTC849 probe targets theclosely related Methylomonas and Methylomarinum clades.The MTC849 oligonucleotide was double-labeled withAtto594 dye (Biomers, Ulm, Germany), and applied to 8 μmsections of sponge tissue using hybridization buffer with 20%formamide as described previously [37]. These hybridizationconditions are assumed to ensure specificity, given the threemismatches that the MTC849 probe had to the 16S rRNAgene sequences of all other Campeche sponge bacteria [38].The general bacterial probe EUB338 [39] was used as apositive control and the NON338 probe was used as a controlfor background autofluorescence [40]. Photomicrographswere acquired with a Zeiss Axioplan 2 epifluorescencemicroscope (Zeiss, Jena, Germany) or with a confocal laser-scanning microscope (LSM 780, Carl Zeiss, Germany).Brightness and contrast of the images were adjusted withAdobe Photoshop (Adobe Systems, Inc., USA).
Transmission electron microscopy (TEM)
Subsamples of the two encrusting and one branchingsponge individuals were washed with 50 mM cacodylatebuffer, post-fixed with 2% osmium tetroxide in the bufferfor 1.5 h at 4 °C, and then washed with Milli-Q water.Following overnight staining in 0.5% uranyl acetate(Merck, Germany), the samples were washed with Milli-Qwater, dehydrated in a graded ethanol series, and thentransferred into propylene oxide (Sigma-Aldrich, Germany).The samples were infiltrated with Epon-812 resin (1:1 resinto propylene oxide) overnight and rinsed in this resin twicefor 2 h. They were then transferred into fresh resin for 1 hand polymerized in embedding capsules at 60 °C for at least48 h. Ultrathin (70 nm) sections were cut with an ultra-microtome (Leica EM UC7, Austria), mounted on piolo-form coated grids, and contrasted with 2.5% uranyl acetatein ethanol for 20 min and subsequently, with Reynold’s leadcitrate for 10 min. Ultrathin sections were imaged at 80 kVon a Tecnai G2 Spirit BioTwin transmission electronmicroscope (FEI Company, USA). We were able to producehigh-quality micrographs of the two encrusting spongeindividuals, but not of the branching sponge individual.
DNA and RNA extraction and sequencing
We extracted DNA and RNA from the two encrustingsponge individuals and from two different branches of the
Mictlan Knoll
500 km
10 km
a21°
20'N
21° 40
'N
22
° 00'N
22°
20'N
93° 50'W 93° 30'W 93° 50'W
20°N
25°
N
30°N
100°W 95°W 90°N 85°W 80°W
Chapopote Knoll
b
I. methanophila
H. (S.) methanophila
Fig. 1 The encrusting sponge Hymedesmia (Stylopus) methanophilasp. nov. and the branching sponge Iophon methanophila sp. nov.colonize asphalt seeps at Campeche Knolls. a Geographic location of
Campeche Knolls and the sponge collection sites, Chapopote andMictlan. b MARUM-QUEST ROV image of H. (S.) methanophila andI. methanophila at Chapopote. Galatheid crabs graze on the sponges
Fueled by methane: deep-sea sponges from asphalt seeps gain their nutrition from methane-oxidizing. . . 1211
same branching sponge individual. DNA and RNA wereextracted in parallel with the AllPrep DNA/RNA Mini Kit(Qiagen, Hilden, Germany) according to the manufacturer’srecommendations, with an extra DNase I digestion step onRNA columns to eliminate contaminating DNA. DNA/RNA quality was assessed with the Agilent 2100 Bioana-lyzer (Agilent, Santa Clara, USA). We were not able toextract RNA in sufficient amounts for transcriptomic ana-lyses of the branching sponge individual. cDNA was syn-thesized with Ovation RNA-Seq System V2 (NuGen, SanCarlos, CA, USA). Genomic DNA and cDNA libraries weregenerated with the DNA library prep kit for Illumina(BioLABS, Frankfurt am Main, Germany). All sampleswere sequenced on the Illumina HiSeq 2500 platform at theMax Planck Genome Centre (Cologne). For one subsampleof the branching sponge, 41 million 150 bp paired-endmetagenomic reads were generated. For the remaining threelibraries, 12.5 million 250 bp paired-end metagenomic readswere generated, while the remaining 15.5–29.5 million weregenerated as 150 bp paired-end reads. In total, 30 and 38million 100 bp paired-end cDNA reads were generated fromthe Mictlan and Chapopote encrusting sponge RNAextracts, respectively.
Genome analyses
Individual metagenomes were assembled with IDBA-UD[41], following decontamination, quality filtering (QV=2) and adapter-trimming with the BBDuk tool from theBBMap suite (Bushnell B, http://sourceforge.net/projects/bbmap/). Individual symbiont genomes were binned basedon genome coverage, GC content and taxonomic affilia-tion using gbtools (Supplementary Figure S1) [42]. Thesebins were reassembled with Spades V3.10 [43, 44],using a maximum k-mer length of 127, following re-mapping of Illumina reads to the bins using BBMap with0.98 minimum identity. Following the manual removal ofcontigs shorter than 800 bp and contamination screening,quality metrics were calculated with Quast [45] andCheckM [46]. Symbiont draft genomes and transcriptomereads were deposited in NCBI under accessionnumbers PRJNA475438 and PRJNA475442. Downstreamgenome analyses are summarized in SupplementaryMethods S1.
Transcriptome analyses
Adapters and ribosomal RNA genes were removed fromtranscriptome reads with BBDuk. Transcriptome reads weremapped to the individual methanotroph genome assemblieswith BBMap (minimum identity value of 0.98). Mapped readswere assigned to genomic features with featureCounts [47].Relative RNA levels were estimated with transcripts per
million (TPM) normalization [48]. The metatranscriptomeswere assembled with Trinity [49] and the transcripts werequantified with an align_and_estimate_abundance.pl scriptfrom the Trinity package, using RSEM quantificationmethod [50].
Phylogenomics
Phylogenomic reconstructions were performed withphylogenomics-tools scripts (https://doi.org/10.5281/zenodo.46122). Marker proteins that are universally con-served across the bacterial domain were extracted fromgenomes using AMPHORA2 [51]. Eighteen single-copymarkers that were present in all genomes analyzed in thisstudy were used for alignment with MUSCLE [52]. Themarker alignments were concatenated into a single parti-tioned alignment. Poorly aligned or misaligned regionswere removed from the alignments. The maximum like-lihood tree was calculated with MEGA7 [53] using the LGmodel [54].
Lipid biomarker analysis
Lipid biomarkers were extracted from one individual ofbranching and one individual of encrusting sponge (bothcollected at Chapopote). Phospholipid fatty acids werehydrolyzed by saponification with 6% KOH in methanol.Sterols were extracted with dichloromethane:methanol (3:1)three times. The resulting total lipid extract and the sapo-nification extract were combined. The hydrolyzed free fattyacid salts were released from the aqueous phase by addingHCl until pH 2. Then, the combined extracts were separatedby solid phase column chromatography into four fractions.The resulting fatty acids and the alcohols were analyzed on aThermo Electron Trace DSQ II coupled gas-chromatograph-mass spectrometer (GC-MS) for quantification and identifi-cation. The GC-MS was equipped with a 30 m HP-5 MS UIfused silica capillary column (0.25 mm i.d., 0.25 µm filmthickness). The carrier gas was helium. The GC temperatureprogram used for both fractions was as follows: 60 °C (1min), from 60 to 150 °C at 10°/min, from 150 to 325 °C at 4°C/min, 25 min isothermal. Identification of compounds wasbased on retention times and published mass spectral data.Double bond positions of unsaturated fatty acids wereidentified by the addition of a dimethyl disulfide (DMDS)adduct to an aliquot of the saponified fatty acid fraction [55].Further, compound-specific carbon stable isotope composi-tions of fatty acids and sterols were measured on a gaschromatograph (Agilent 6890) coupled with a Thermo Fin-nigan Combustion III interface to a Finnigan Delta Plus XLisotope ratio mass spectrometer (GC-IRMS). The GC con-ditions were identical to those mentioned above for GC-MSanalyses.
1212 M. Rubin-Blum et al.
Results and Discussion
Novel sponge species are abundant at Chapopote
Morphological analyses of both sponge morphotypesrevealed that these have not yet been described. We namethem here Hymedesmia (Stylopus) methanophila sp. nov.and Iophon methanophila sp. nov. (order Poecilosclerida).These belong to two different genera and are only distantlyrelated to each other (based on 90% identity of their cyto-chrome c oxidase subunit I (COI) gene sequences). A fulldescription of the morphology and phylogeny of these twosponge species is available in Supplementary File SF1.
At Chapopote and Mictlan, H. (S.) methanophilaand I. methanophila occurred on fragmented asphalt
accumulations next to sites of active seepage, characterizedby gas and oil effusions and darkly colored substrates(Supplementary Figure S2). I. methanophila individualswere often observed growing on the tubes of tubeworms.Based on our mosaic mapping of the Chapopote site, weestimate that the sponges colonized ~30% of the hardsubstrates in areas of active seepage (Supplementary Fig-ure S2). We often observed grazers in association with thesponges, in particular, galatheid crabs (Fig. 1). Theseobservations, together with observations of similar spongeabundances during a 2005 cruise to Chapopote [56], suggestthat these sponges have made up a considerable part of thebiomass at Chapopote for over twelve years, and are thuslikely to influence the composition of the seep communityand contribute to its food web.
Table 1 Relative abundance and phylogenetic assignment of sponge-associated bacteria based on metagenomic and metatranscriptomic coverageof their 16S rRNA gene sequences and the coverage of metagenome assembled genomes (MAGs)
16S rRNA gene coverage egarevocGAMegareva
ANRANDHmetMict
(2.7) HmetChap
(1.0) ImetChap1
(0.4) ImetChap2
(0.8) HmetMict
(1.6) HmetChap
(5.6) HmetMict HmetChap ImetChap1 ImetChap2
MOX (MMG2, Proteobacteria) 58.8 50.4 58.1 52.5 42.8 37.6 67.6 52.1 45.3 35.9
Thioglobaceae - SUP05 (Proteobacteria)
5.7 6.9 12.4 14.8 6.8 7.3 4.6 5.2 17.0 20.2
Cycloclasticus (Proteobacteria) 10.8 6.2 6.5 5.1 8.9 1.8 7.9 8.4 6.7 6.0
E01-9C-26 marine groupa
(Proteobacteria) 6.3 14.8 0.6 0 10.0 10.0 14.8 21.6 0 0
JTB23 (Proteobacteria) 10.9 11.9 10.7 12.9 8.5 5.3 1.4 4.3 15.1 18.7
Nitrosopumilus (Thaumarchaeota)
0.1 3.2 0 0.6 8.1 28.7 0.4 6.2 0 0.9
Aqs2 Betaproteobacteriab
(Proteobacteria) 0 0 6.8 7.5 0 0 0 0 8.6 9.0
Cytophagales (Bacteroidetes) 0 0 2.6 3.8 0 0 0 0 3.2 4.6
BD7-8 marine group (Proteobacteria)
3.4 2.9 0 0 7.6 3.0 2.1 1.4 0 0
Endozoicomonas (Proteobacteria) 2.1 1.2 3.2 2.1 3.2 2.1 0 0 0 0
KI89A clade ((Proteobacteria)) 1.7 1.0 0.3 0.3 3.1 0.9 1.1 0.7 0.4 0.5
Azospirillum (Proteobacteria) 0.2 1.0 0 0 0.9 2.6 0 0 0 0
Nitrospina (Nitrospinae) 0 0.4 0 0 0.1 0.6 0 0 0 0
OCS116 (Proteobacteria)) 0 0 1.5 1.3 0 0 0 0 2.0 2.1
SAR324 (Proteobacteria)) 0 0 0.5 0.9 0 0 0 0 0.8 1.3
Spirochaetales (Spirochaetes) 0 0 0.3 0.2 0 0 0 0 0.4 0.5
Bdellovibrio (Proteobacteria) 0 0 0.3 0.2 0 0 0 0 0.3 0.2
Color code (%) ≤0 ≤10 ≤30 ≤60
Classifications at the Phylum level are mentioned in parentheses. The 16S rRNA sequences of the MOX symbionts were identical within each hostspecies, and differed slightly, but distinctly between the two host species (97.8% identity). The following names are abbreviated: H. (S.)methanophila from Mictlan (HmetMict) and Chapopote (HmetChap); I. methanophila from Chapopote subsamples 1 and 2 (ImetChap1 andImetChap2); methane-oxidizing bacteria (MOX); marine methylotrophic group 2 (MMG2). Numbers in parentheses within the column headersrepresent the ratio between the MOX 16S and the host’s 18S rRNA gene coverageaE01-9C-26 marine group is a monophyletic ‘sponge-enriched’ gammaproteobacterial cladebAqs2 symbiont is a monophyletic ‘sponge-enriched’ clade of Proteobacteria, which includes symbionts of the low microbial abundance spongesAmphimedon queenslandica and Crambe crambe
Fueled by methane: deep-sea sponges from asphalt seeps gain their nutrition from methane-oxidizing. . . 1213
Campeche sponges host high abundances ofmethane-oxidizing bacteria
Analyses of the microbial communities hosted by H. (S.)methanophila and I. methanophila revealed that theirdominant members were MOX (35.9–67.6% of the micro-bial community; these values and the following in thisparagraph are based on relative 16S rRNA read frequenciesin the metagenomes, as well as the frequencies of metage-nomic reads that mapped to the MOX genomes, see Sup-plementary Methods and Table 1). In both host species,Proteobacteria made up the vast majority of the microbialcommunity (94.8–99.9%), and included chemoautotrophicsulfur-oxidizing bacteria from the SUP05 clade (4.6–20.2%) and hydrocarbon-degrading Cycloclasticus sym-bionts (6.0–8.4%), which have been described elsewhere[33]. Such high abundances of Proteobacteria have not beencommonly found in HMA sponges. Most HMA sponges
harbor highly diverse microbial communities that are oftendominated by the bacterial phyla Chloroflexi, Acid-obacteria, and Actinobacteria [57].
Although the low phylogenetic diversity of the microbialcommunity in Campeche sponges is typical of LMA hosts,fluorescence in situ hybridization (FISH) revealed that theabundances of bacteria were more typical of HMA sponges[5] (Fig. 2). FISH with a probe specific to the MOX 16SrRNA sequences in both host species revealed that thesebacteria were present in high numbers in the spongemesohyl (the gelatinous matrix within a sponge that fills thespace between its external and internal cell layers).Although the Campeche sponges belong to an order (Poe-cilosclerida) previously described as consisting of onlyLMA species, they have traits typical of both LMA sponges(low phylogenetic diversity) and HMA sponges (highsymbiont abundances). They thus represent an exception tothe described dichotomy between HMA and LMA sponges.
0.5 mm
e1
e2
e
e
b c
0.2 mm
a
d
e
10 μm
200 μm
e1
e2
50 μm
e
594, MOX
DAPI, DNA
Fig. 2 Fluorescence in situ hybridization (FISH) images of the symbioticmethane-oxidizing bacteria (MOX) in Campeche sponges. FISH wasperformed with a probe specific to the symbiotic MOX on 10 μm thicksections of I. methanophila and H. (S.) methanophila. Colors: MOXsymbionts, magenta; DNA (DAPI staining), blue; autofluorescence at~520 nm with ~490 nm excitation (FITC filter), green. a Overview of I.methanophila. The image is a mosaic of five aligned images (lines markborders between the images). b Overview of H. (S.) methanophila. Theimage is a mosaic of twenty aligned images (lines mark borders betweenthe images, e: embryos in various developmental stages. c MOX
symbionts are abundant in the mesohyl of I. methanophila. d Detail ofembryos labeled e1 and e2 in panel (b). e FISH images of embryolabeled e2 in panel (b) using gray intensity representation to distinguishFISH signal of MOX symbionts (top panel) from DAPI staining (middlepanel) and autofluorescence (bottom panel). Images from the mosaics,additional images of MOX in tissues of H. (S.) methanophila embryosand 3-dimensional z-stack reconstructions are available at https://figshare.com/projects/Fueled_by_methane_Deep-sea_sponges_from_asphalt_seeps_gain_their_nutrition_from_methane-oxidizing_symbionts/23020
1214 M. Rubin-Blum et al.
Transmission electron microscopy (TEM) of H. (S.)methanophila revealed that their mesohyl was colonized bya bacterial morphotype characterized by (i) a coccoid shapeand intracytoplasmic membrane stacks typical of MOXbacteria; (ii) sizes of 500–1300 nm length and 500–900 nmwidth, and (iii) electron-lucent granules typical of storagecompounds, most likely glycogen (Fig. 3a, SupplementaryNote 1). The distribution pattern of these morphotypes wassimilar to that of the MOX bacteria identified with FISH,indicating that we could identify the MOX symbiont of H.(S.) methanophila with TEM based on its characteristicultrastructure. We regularly observed MOX in stages ofdivision, indicating that they were actively growing in thesponge matrix (Fig. 3b). The MOX were generally extra-cellular, but we occasionally found them within hostvacuoles in different stages of degradation (Fig. 3c), indi-cating that the sponge cells engulf and digest these sym-bionts. Bacteria with a very different morphology than thatof the MOX symbionts occurred in what looked like spe-cialized bacteriocyte cells described from other spongespecies [58, 59], but we never observed MOX in these hostcells (Fig. 3f). The symbiotic MOX were most abundant inmesohyl regions near the choanocyte chambers (Fig. 3a).This indicates that the symbiotic MOX benefit from beingclose to the flagellated choanocytes, most likely because
these host cells pump the methane- and oxygen-containingseawater from the seep environment into the sponge matrix,and thus provide the symbionts with the reductants andoxidants they need for their energy and carbon metabolism.
Specificity of the sponge-MOX association
The comparison of the MOX symbiont genomes from thetwo Campeche sponge species suggests that these are spe-cific to their hosts. The two H. (S.) methanophila indivi-duals that were collected 25 km apart at Chapopote andMictlan, hosted nearly identical MOX genotypes (99.6 ±0.9% average nucleotide identity (ANI)). Only one I.methanophila individual could be collected from the Cam-peche seeps (at Chapopote), but ANI values of the MOXsymbiont genomes from two different branches of thisindividual were nearly identical (100 ± 0.6%). Comparisonof the MOX symbiont genomes from the sympatric H. (S.)methanophila and I. methanophila individuals, which werecollected from the same asphalt piece at Chapopote,revealed that these differed considerably from each other(93.6 ± 2.5% ANI), and can be considered to belong to twodifferent species (ANI < 95%, [59]). These results suggestthat each sponge species hosts a specific MOX genotype,and imply the presence of recognition and selection
l
bc
n
MOX
MOX
he
5 μm 300 nm 200 nm
1 μm 1 μm 5 μm
a b c
d e f
b
c
bc
embryo
MOX
glycogen
n
sp
sp
spsp
bcMOX
MOX
sp
MOX
Fig. 3 Transmission electron microscopy images of H. (S.) methano-phila. a Symbiotic methane-oxidizing bacteria (MOX) are abundant inthe mesohyl, particularly in regions close to the choanocyte chambers(chambers not visible in image), sp= sponge cells, n= nucleus.b High-resolution image of white box labeled b from panel (a),showing intracellular membranes (arrows) typical for MOX. TheMOX at the lower half of the image are dividing; bc= bacteria with a
different morphology than the MOX symbionts. c MOX symbiont in alysosome of an amebocyte, l= lysosome. d Mesohyl with an embryosurrounded by symbiotic MOX, and a bacteriocyte containing bacteriawith a different morphology than the symbiotic MOX (bc). e MOXsymbionts within an embryo. he= heterogeneous yolk. f A bacter-iocyte containing bacteria with a different morphology than the MOXsymbionts. bc= bacteria; n= nucleus
Fueled by methane: deep-sea sponges from asphalt seeps gain their nutrition from methane-oxidizing. . . 1215
mechanisms that underlie the potential specificity of thissymbiosis.
The symbiotic MOX belong to the MarineMethylotrophic Group 2 clade
Phylogenomic reconstruction of eighteen marker proteins,as well as the phylogenies of 16S rRNA and pmoA (parti-culate methane monooxygenase) genes, revealed that thesponge-associated MOX belong to the Marine Methylo-trophic Group (MMG) 2 clade (also known as deep-seaclade 2) within the family ‘Methylomonadaceae’, orderMethylococcales (Fig. 4, Supplementary Figures S3 and S4)[61]. To date, there are no pure cultures of bacteria from theMMG2 clade, but recently the genomes of two MMG2MOX enriched from North Sea sandy sediments withmethane as the sole carbon and energy source weresequenced [62]. Together with the Campeche MOX
symbionts, these are currently the only genomes that havebeen sequenced from the MMG2 clade.
Interestingly, MMG2 bacteria have been reported fromother marine hosts based on 16S rRNA and pmoA genesequencing, although their abundances appear to generallybe considerably lower than in the Campeche sponges(Supplementary Figures S3 and S4). MMG2-relatedsequences were described from ciliates collected atmethane seeps along the eastern Pacific coast (1.7–9.7% ofthe ciliate bacterial community based on 16S rRNA tagsequencing) [63], and the squat lobster Shinkaia crosnierifrom hydrothermal vents off Japan (1.7–12.0% of 16SrRNA clones), [64]. MMG2 sequences were also found intwo sponge species: (i) an unidentified poeciloscleridsponge from seeps in the Gulf of Mexico that is morpho-logically distinct from our samples [28], and (ii) Cladorhizamethanophila from mud volcanoes off Barbados (16–25%based on 16S rRNA tag sequencing [25]). MOX belonging
Fig. 4 Phylogenomic tree and metabolic repertoire of the sponge MOXsymbionts and related bacteria (45 sequences total). The sponge MOXsymbionts and the two free-living MOX from North Sea sedimentenrichments (provided by B. Vekeman) are currently the only genomesavailable for the MMG (Marine Methylotrophic Group) 2, although16S rRNA data indicates that they are widespread (see SupplementaryFigure S3 16S rRNA tree). Eighteen single-copy markers as defined inthe AMPHORA core bacterial phylogenetic marker database were
used in the analysis. The tree is drawn to scale, with branch lengthsrepresenting the number of substitutions per site. The percentage oftrees in which the associated taxa clustered together was determinedbased on 100 bootstrap resamples. The analysis included 3706 posi-tions. *These clades were formerly included in the family Methylo-coccaceae, and recently placed together with the MMG1 and MMG2clades within the family ‘Methylomonadaceae’ (order Methylo-coccales), based on the Genome Taxonomy Database
1216 M. Rubin-Blum et al.
to the MMG2 clade were also found in a Sibolginum cf.poseidoni tubeworm from a mud volcano in the NortheastAtlantic (the nine 16S rRNA clones sequenced from thetubeworm all belonged to MMG2 [65]).
Common to all these hosts is that they occur in methane-rich environments, such as mud volcanoes, seeps, andhydrothermal vents, where free-living MMG2 bacteria areknown to be abundant in both the sediment and the water
column [36]. Given the broad range of hosts associated withbacteria from the MMG2 clade, from protists to animalsfrom different phyla, it is highly likely that these associa-tions were established multiple times independently of eachother. Moreover, the MOX bacterial partners in these asso-ciations are phylogenetically distinct from those of the deep-sea bathymodiolin mussels, which belong to the MMG1clade (Fig. 4 and Supplementary Figures S3 and S4) [66].
a
b
Fueled by methane: deep-sea sponges from asphalt seeps gain their nutrition from methane-oxidizing. . . 1217
Our study thus contributes to revealing the previouslyunrecognized diversity of methanotrophic associations onboth the bacterial and host side. This diversity suggests thatthere are strong selective advantages for both partners inestablishing these beneficial associations in methane-richenvironments. Revealing the factors that contribute to theseselective advantages, and distinguishing these from themechanisms behind the broad host promiscuity in MMG1MOX and the apparent specificity of MMG2 MOX tobathymodiolin mussels, will contribute to understanding theecological and evolutionary requirements for establishingthese symbioses.
The symbiotic MOX may be vertically transmitted
We found evidence for vertical transmission of the sym-biotic MOX from one generation to the next. Embryos inthe matrix of both H. (S.) methanophila individuals con-tained MOX symbionts based on FISH and TEM analyses(Figs. 2 and 3; Supplementary Figure S5). The symbioticMOX were observed in the matrix between the embryocells, but never inside of sponge cells (Fig. 3e). MOX werealso found between follicle-like cells surrounding the egg,as well as on the periphery of follicle-like cells (Supple-mentary Figure S5). In the seep sponge Cladorhizamethanophila, bacteria with features typical of methane-oxidizing bacteria were also observed in embryos with
TEM, indicating that symbionts are transmitted vertically inthis sponge species as well [67].
Genome reduction in the symbiotic MOX
Additional support for our conclusion that the Campechesponge MOX symbionts are vertically transmitted is pro-vided by the comparison of their genome sizes to those offree-living MOX bacteria. The estimated genome sizes ofthe Campeche MOX symbionts were between 2.0 and 2.2Mbp, and thus considerably reduced in comparison to the3.9 and 4.0 Mbp genomes of their close relatives fromNorth Sea sediment enrichments (Fig. 5a, SupplementaryTable S1). The genomes of the sponge MOX symbiontswere also reduced in comparison to those of cultivated‘Methylomonadaceae’ (4.5–5.2 Mbp), although some free-living ‘Methylomonadaceae’ may have similarly smallgenomes, e.g., Methylococcales bacterium OPU3_G-D_OMZ from a marine metagenome [68]. Given that thesponge-associated MOX genomes were highly complete(94.6–96.5%), it is unlikely that we underestimated theirsizes due to incomplete binning (Supplementary Table S1).As genome reduction is typical for vertically transmittedsymbionts, the small genome sizes of the Campeche spongeMOX symbionts may be a result of accelerated genomeevolution through vertical transmission [69–71]. The verylow guanine+ cytosine (GC) content of the symbiontgenomes (37.7 ± 0.1% versus 51 ± 4% in cultivated‘Methylomonadaceae’) may also be a result of verticaltransmission, as known from many other vertically trans-mitted symbionts Fig. 5a) [72]. However, the closest free-living relatives of the sponge symbionts, the North Seabacteria from sediment enrichments (Fig. 4), have similarlylow GC contents of 37.7%, indicating that other factors,such as energetic constraints, may have played a role in thegenome evolution of this clade [73, 74].
Optimization of methane assimilation in spongesymbionts
To better understand the functional adaptations of thesponge MOX to their symbiotic lifestyle, we compared theirgenomes to those of free-living MOX within the ‘Methy-lomonadaceae’ as well as the MOX symbionts of bath-ymodiolin mussels (Figs. 4 and 5). We identified 1050genes in the core genome of the free-living ‘Methylomo-nadaceae’, 163 of which were not found in the spongeMOX symbionts (Fig. 5a, Supplementary Table 3a). Com-parative genomics revealed that the functional repertoire ofthe sponge MOX symbionts for the assimilation of methaneand other carbon compounds was reduced in comparison tofree-living bacteria within the ‘Methylomonadaceae’(Figs. 4, 5b, Supplementary Figure S6).
Fig. 5 Comparison of genomes from the sponge symbionts and othergammaproteobacterial methane-oxidizing bacteria (MOX). a Dis-tribution of estimated genome sizes, guanine-cytosine (GC) contentand the estimated number of toxin-antitoxin components in genomesof methane-oxidizing bacteria (MOX). The blue and red lines showpan genomes (blue) versus core genome (red) development plots(obtained by iteratively adding one genome at a time to the comparisonin the defined order (starting with the first genome from the left).Comparative genomic analysis led to a pan-genome estimate of 16476coding sequences, of which 1050 formed the core genome. The gra-dual change in the slope of the pan-genome development curve for theMMG 2 symbionts suggests that this clade is sufficiently sampled. 163genes were subtracted from the core genome of ‘Methylomonadaceae’after addition of the sponge symbionts. Mbp=million base pairs.b Principal component analysis based on the relative abundance ofclusters of orthologous groups (COGs) encoded by symbiotic and free-living MOX. The following COG abbreviations are shown: [D] cellcycle control, cell division, chromosome partitioning, [M] cell wall/membrane/envelope biogenesis, [N] cell motility, [O] post-translational modification, protein turnover, and chaperones, [T] sig-nal transduction mechanisms, [U] intracellular trafficking, secretion,and vesicular transport, [V] defense mechanisms, [A] RNA processingand modification, [J] translation, ribosomal structure and biogenesis,[K] transcription, [L] replication, recombination and repair, [C] energyproduction and conversion, [E] amino acid transport and metabolism,[F] nucleotide transport and metabolism, [G] carbohydrate transportand metabolism, [H] coenzyme transport and metabolism, [I] lipidtransport and metabolism, [P] inorganic ion transport and metabolism,[Q] secondary metabolites biosynthesis, transport, and catabolism, [S]function unknown
1218 M. Rubin-Blum et al.
Similar to other ‘Methylomonadaceae’, the sponge MOXsymbionts are type I methanotrophs: The ribulose mono-phosphate (RuMP) cycle appears to be their sole pathwayfor methane incorporation, as key enzymes for the assim-ilatory serine cycle, such as hydroxypyruvate reductase,were not found [75]. Genes not found in the sponge MOXincluded those encoding the Entner–Doudoroff (EDD)pathway enzymes, phosphogluconate dehydratase (Edd)and 2-dehydro-3-deoxy-phosphogluconate aldolase (Eda),which play a role in methane carbon assimilation via theRuMP pathway [76]. The sponge MOX symbionts appearto assimilate single carbon compounds only via the highlyefficient Embden–Meyerhof–Parnas (EMP) variant of theRuMP cycle, based on the presence and substantialexpression of the genes encoding fructose-bisphosphatealdolase (fba) and pyrophosphate-dependent phospho-fructokinase (pfk) [76] (Fig. 6). The sponge MOX sym-bionts may, therefore, be more efficient in using single
carbon compounds than the methane-oxidizing symbiontsof bathymodiolin mussels, which only employ the lessefficient EDD variant of the RuMP pathway [76, 77]. Thissuggests that the symbiotic MOX of sponges are able toprovide nutrition to their hosts at lower methane con-centrations than the mussel symbionts. This hypothesis issupported by our mapping analyses of the seafloor atChapopote, as sponges were often situated further fromhotspots of active gas and oil seepage than the mussels(Supplementary Figure S2).
The symbiotic MOX appear to use a minimal suite ofenzymes needed for methane assimilation. As opposed tomost sequenced, free-living MOX, in which two variants ofmethanol dehydrogenase, the lanthanide-dependent (XoxF)and the calcium-dependent (MxaF) methanol dehydrogenaseco-occur, only XoxF was found in the sponge symbionts(Fig. 4). Furthermore, only one pmoCAB operon encodedthe methane monooxygenase, while the pxmABC operon
Fueled by methane: deep-sea sponges from asphalt seeps gain their nutrition from methane-oxidizing. . . 1219
was not found (Fig. 4). Despite this genetic minimalism ingenes involved in methane oxidation, these genes, togetherwith the RuMP pathway, were highly expressed by thesymbionts (Fig. 6). Moreover, these genes were among thetop 1% of the most well-expressed transcripts in the meta-transcriptomes, which included reads that mapped to boththe host and the symbionts. This suggests that methaneassimilation by the symbiotic MOX was among the mostactive metabolic processes in the sponge holobiont.
The sponge MOX symbionts may be able to use othersources of carbon besides methane. Their genomes
contained the genes for the TCA cycle, glycogen synthesisand degradation, as well as the import of multicarbon sub-strates via C4-dicarboxylate tripartite ATP-independentperiplasmic (TRAP) transporters (Fig. 6, SupplementaryNote 1). Moreover, most of these genes were expressed(Fig. 6). This suggests that similar to the mussel MOXsymbionts, which also may use other carbon sources besidesmethane [77], the sponge MOX symbiont have evolvedmechanisms to deal with fluctuations in methane availability.Indeed, our TEM observations of electron-lucent granulestypical of carbon storage compounds in the MOX spongesymbionts supports our assumption that glycogen may beused to buffer against periods of methane deprivation.
Reductive evolution and adaptation to thesymbiosis
Comparative analyses of clusters of orthologous groups(COGs) in the sponge MOX with those of free-living‘Methylomonadaceae’, as well as the MOX symbionts ofbathymodiolin mussels revealed that physical and bioticinteractions with the environment appear to have shaped thepan-genome of free-living MOX. In these MOX, functionssuch as “cell motility”, “inorganic ion transport and meta-bolism”, “signal transduction mechanisms”, “membranebiogenesis”, and “function unknown” (this diverse groupincludes toxins, type VI secretion system, antibiotic resis-tance, phage and plasmid proteins) were enriched in com-parison to the sponge and mussel symbionts (Fig. 5b). TheCOGs “translation, ribosomal structure and biogenesis”,“coenzyme transport and metabolism” and “amino acidtransport and metabolism” were enriched in the spongeMOX symbionts, suggesting that their metabolic repertoirebeyond these housekeeping functions is limited. Largenumbers of transposases and integrases resulted in theenrichment of the COG “replication, recombination andrepair” in the MMG1 clade, which comprises symbionts ofbathymodiolin mussels and a single free-living bacteriumM. sedimenti. Expansion of transposable elements is morecommon in symbionts that have recently transitioned to anobligate, host-associated lifestyle, than in symbionts thathave associated with their hosts over long evolutionary timeperiods [68, 69, 78, 79]. The sponge symbioses, therefore,may have pre-dated those of mussels. However, this ishighly speculative, as the free-living MOX M. sedimentialso has high numbers of transposable elements. The onlygenes that were shared between the sponge and musselsymbionts, yet not present in most free-living ‘Methylo-monadaceae’, were the narGHIJ genes, encoding enzymesfor nitrate respiration (Supplementary Note 2). Thus, thereappears to be little convergence in the mechanisms thesponge and mussel symbionts have evolved to establish andmaintain associations with their hosts.
Fig. 6 Central carbon and nitrogen metabolism in the sponge MOXsymbionts. The reconstruction is based on the genomes of I. metha-nophila and H. (S.) methanophila symbionts and on the averageexpression in the two transcriptomes of H. (S.) methanophila sym-bionts. Boxes represent enzyme subunits and the abbreviationsrepresent the genes that encode the respective subunit. Boxes arecolored according to the expression value of the gene. Thefollowing genes are abbreviated: pmoABC, particulate methanemonooxygenase subunits A, B and C; xoxF, methanol dehydrogenase;fae, formaldehyde activating enzyme; mtdB, methylene tetra-hydromethanopterin dehydrogenase; mch, methenyltetrahy-dromethanopterin cyclohydrolase; fhcABCD, formyltransferase/hydrolase complex; fdhAB, formate dehydrogenase subunits alpha andbeta; mtdA, methylene tetrahydrofolate/methylene tetra-hydromethanopterin dehydrogenase; ftfL, formate-tetrahydrofolateligase; hps, 3-hexulose-6-phosphate synthase; hpi, 6-phospho-3-hex-uloisomerase; pfk, pyrophosphate-dependent phosphofructokinase;fbp, fructose-1,6-bisphosphatase; fba, fructose-bisphosphate aldolase;tpi, triosephosphate isomerase; tkt, transketolase; talB, transaldolase;rpe, ribulose-phosphate 3-epimerase; rpiA, ribose-5-phosphate iso-merase; gpi, glucose-6-phosphate isomerase; zwf, glucose-6-phosphate1-dehydrogenase; pgl, 6-phosphogluconolactonase; gndA,6-phosphogluconate dehydrogenase; gapdh, glyceraldehyde3-phosphate dehydrogenase; pgk, phosphoglycerate kinase; gpml, 2,3-bisphosphoglycerate-independent phosphoglycerate mutase; eno,enolase; pyk, pyruvate kinase; por, pyruvate-flavodoxin oxidor-eductase; dlat, acetyltransferase component of pyruvate dehy-drogenase complex; dld; dihydrolipoamide dehydrogenase of pyruvateor 2-oxoglutarate dehydrogenase complexes; pdhA1, pyruvate dehy-drogenase E1 component subunit alpha; oadABG, oxaloacetate dec-arboxylase, alpha, beta and gamma chains; pgm, phosphoglucomutase;glgC, glucose-1-phosphate adenylyltransferase; glgA, glycogen syn-thase; glgB, 1,4-alpha-glucan branching enzyme; smht, serine hydro-xymethyltransferase; agt/sgaA, serine-glyoxylate aminotransferase;mclA, malyl-CoA lyase; mtkAB, malate thiokinase, alpha and betasubunits; mdh, malate dehydrogenase; gltA, citrate synthase; acn,aconitase; idh, isocitrate dehydrogenase; sucA, 2-oxoglutarate dehy-drogenase E1 component; dlst, dihydrolipoyllysine-residue succinyl-transferase component of 2-oxoglutarate dehydrogenase complex;sucCD, succinate-CoA ligase subunits beta and alpha; sdhABCD,succinate dehydrogenase complex subunits; fumC, fumarate hydratase;dctMPQ, C4-dicarboxylate TRAP transporter subunits; gltDB, gluta-mate synthase, large and small chains; glnA, glutamine synthetase;amt, ammonium transporter; nrt, nitrate transporter; nirBD, assim-ilatory nitrite reductase small and large subunits; narGHIJ, respiratorynitrate reductase, alpha-gamma subunits; urtABCDE, urea ABCtransport system subunits; ureABCDEFG, urease subunits; atpABC-DEFGH, subunits of the membrane-bound ATP synthase; cox123,cytochrome c oxidase subunit I-III; cytc1, cytochrome c-1, cytochromeb-c1 complex; urc1, cytochrome b-c1 complex subunit 1; cytb, cyto-chrome b, cytochrome b-c1 complex
1220 M. Rubin-Blum et al.
Analysis of the 163 core proteins that were unique to thefree-living ‘Methylomonadaceae’ revealed that the spongesymbionts appear to lack the following functions: (i)nutrient uptake (nitrate ABC transporter NrtABC, the Phoregulon and phosphate transport system); (ii) secondarymetabolite production and secretion (Rml proteins thatcatalyze synthesis of nucleotide/polyketide sugars and Gspbuilding blocks of the type 2 secretion system); and (iii)adaptation to hypoxia, based on the absence of the high-affinity ba3-type cytochrome oxidase [80] (Fig. 4, Supple-mentary Table 3a). Further indications that the spongesymbionts are not well adapted to hypoxia was the apparentlack of genes encoding the soluble NAD-reducing hydro-genase and acetate kinase, which catalyze fermentation ofmethane-derived products in many other MOX [76](Fig. 4). Some genes and pathways involved in nitrogenmetabolism also appear to be lacking in the sponge sym-bionts. Unlike most free-living MOX, the sponge symbiontsmay be incapable of nitrogen fixation and dissimilatorynitrite reduction to nitrous oxide (Fig. 4). As suggestedpreviously for sulfur-oxidizing symbionts [81], a divergentrespiratory nitrate reductase NarGHIJ, uncommon in mostMOX, may play a role in non-canonical nitrate assimilationin the sponge MOX symbionts, which appear to lack theassimilatory nitrate reductase Nas (Supplementary Note 2).Altogether, these findings suggest a reduced functionalrepertoire of the sponge symbionts, which may have beenshaped by (i) adaptation of the sponge MOX to the chemicalenvironment within the host milieu, (ii) metabolic optimi-zation via reduction of functional redundancy, and (iii) avery limited range of biotic interactions with free-livingmicroorganisms. The reduction of such a wide array offunctions that appear to be essential for a free-living life-style suggests that the sponge MOX may have entered the‘rabbit hole’ of obligatory symbiosis, and are no longer ableto actively grow outside of their host.
Functional homogeneity between the symbioticMOX from the two sponge species
Comparative genomics revealed that the MOX symbiontsfrom the two sponge species were similar in most of theircore metabolic pathways. In total, 1042 genes were com-mon to their core genome, while 372 and 357 genes wereunique to H. (S.) methanophila and I. methanophila sym-bionts, respectively (Supplementary Figure S6). The mainfraction of the genomic content unique to each symbiontconsisted of poorly-annotated open reading frames (ORFs),primarily including mobile elements and phage-derivedsequences, often stabilized by toxin-antitoxin systems [82].These toxin-antitoxin systems are abundant in the genomesof all sequenced aerobic methanotrophs and account for asignificant part of their genomic variability (Fig. 5a). Only
three sets of genes with well-described metabolic functionswere present in the genomes of the H. (S.) methanophilaMOX, but appear to be absent in those of the I. methanophilaMOX: (i) genes encoding a urease and urea transporter, (ii) agene encoding a C4-dicarboxylate transporter, and (iii) asecondary metabolite synthesis cluster, most likely an arylpolyene of unknown function [83] (Supplementary Fig-ure S7). No genes with a well-described metabolic functionwere found that were unique to the I. methanophila MOX incomparison to the H. (S.) methanophila MOX.
Eukaryote-like proteins are encoded in the genomesof the sponge MOX
The genomes of the sponge MOX symbionts contained 823genes that were unique to the symbionts from both spongespecies, and not present in free-living relatives (97 of whichbelong the sponge MOX symbiont core genome, see Sup-plementary Note 3 for more details). Some of thesesymbiont-specific sequences were ORFs as large as 14.6 kb,and contained one or more eukaryotic-like domains (ELD),such as leucine-rich repeats, cadherin-like domains, andbacterial immunoglobulin-like domains (SupplementaryTable S3, Supplementary Figure S8). ORFs with multipleELDs are well-known from intracellular bacterial pathogensand have also been regularly found in the bacterial meta-genomes of other sponge species, where they encodeeukaryote-like proteins (ELPs), which are likely involved insymbiotic interactions with their hosts (e.g., ref. [84]). In theMOX symbionts of H. (S) methanophila, some of the ELPscontained autotransporter beta-domains. These are knownfrom many gram-negative bacteria, and encode a proteininvolved in forming a pore through the outer membrane[85]. This suggests that the ELPs could be secreted, pro-viding further evidence for their potential role in symbiont–host interactions. BLAST analysis of the sponge MOXsymbiont ELPs against the NCBI database resulted in besthits to the ELPs in the Cycloclasticus symbionts of theCampeche sponges (e.g., ORU94421.1, 98% identity and78% coverage compared to the 14.6 kb ORF from the H.(S.) methanophila MOX). The similarity of ELPs fromsymbionts that inhabit the same host but belong to verydistant bacterial lineages suggests the convergent evolutionof these putative symbiosis factors.
The symbiotic MOX appear to be the primary sourceof sponge carbon
Given the high abundances of MOX symbionts in bothCampeche sponge species, we hypothesized that they playan important role in their nutrition. To test this hypothesis,we analyzed phospholipid fatty acids (PLFAs) and sterols inthe sponges (see Supplementary Note 4 for more details).
Fueled by methane: deep-sea sponges from asphalt seeps gain their nutrition from methane-oxidizing. . . 1221
Monounsaturated fatty acids known to be specific to MOX[86], comprised 27% of all lipids of H. (S.) methanophilaand 14% of all lipids in I. methanophila individuals, con-firming high abundances of symbiotic MOX in both hostspecies (Table 2). Carbon isotopic signatures of these fattyacids reflected those of Campeche methane (δ13Cmethane wasbetween −50 and −45‰ [32], δ13CSpongeMOXfattyacids wasbetween −51 and −44‰, Supplementary Table S3). Thissuggests that in contrast to cultivated type I methanotrophs[87, 88], the sponge MOX symbionts, as well as the musselsymbionts [86], do not fractionate methane carbon duringthe biosynthesis of their fatty acids, possibly due to periodicmethane limitation.
We were able to trace elongation of the MOX-specificfatty acids n-C16:1ω8 and n-C16:1ω7 to n-C18, n-C20, n-C22, n-C24, n-C26 with the same double bond positions and similarisotopic signatures in H. (S.) methanophila. This indicatesthat carbon compounds from the MOX symbionts wereincorporated into their host’s biomass (Table 2, Supple-mentary Table S3). I. methanophila also showed a similarchain-elongation pattern, however, the double bond posi-tions were not exclusively ω8 and ω7, suggesting a lowerdegree of carbon incorporation from its MOX symbiontsthan in H. (S.) methanophila. Given the very low isotopicvalues of most sponge-derived lipid biomarkers, methaneappears to be the main carbon source of the Campechesponges, and we hypothesize that the majority of themethane-derived carbon in the sponges originates from theirsymbionts (Supplementary Note 5).
Conclusions
Our study revealed that two, only distantly related, speciesof sponges independently established highly specific,nutritional symbioses with two very closely related MOXbacteria. This convergence in symbiont acquisition under-scores the strong selective advantage for these sponges inharboring MOX bacteria in the food-limited deep sea: Theseanimals gain access to a carbon and energy source,
methane, that they cannot access on their own. For thesymbionts, one of the main advantages is that the spongesprovide them with simultaneous and continuous access toboth methane and oxygen via active water pumping andhigh surface to volume ratios. In contrast, free-livingmethanotrophs are limited to a very narrow range of habi-tats in which methane and oxygen co-occur.
Most research on symbioses between MOX bacteria andeukaryotes has focused on MOX from the MMG1 clade thathave, so far, only been found in Bathymodiolus mussels. Incontrast, only little is known about the symbioses betweenMOX from the MMG2 clade and their hosts. The diversityof hosts that these MMG2 MOX are associated with, ran-ging from ciliates to sponges, lobster and tubeworms, isonly beginning to be recognized. Our study provides mul-tifaceted insights into the genomic and metabolic potentialof MMG2 MOX symbionts from deep-sea, seep sponges.Future studies of MMG2 MOX from other host groups willallow comparative analyses of the traits that have enabledthese bacteria to independently colonize eukaryotic hostsmultiple times in convergent evolution.
Acknowledgements The authors thank all individuals who helpedduring the R/V Meteor research cruise M114, including onboardtechnical and scientific personnel, the captain and crew, and the ROVMARUM-Quest team. We thank the Max Planck-Genome-CentreCologne (http://mpgc.mpipz.mpg.de/home/) for generating the meta-genomic and the metatranscriptomic data used in this study, theImaging Core Facility at the University of Würzburg, Germany forembedding of the TEM samples, the Central Microscopy unit at theUniversity of Kiel, Germany for access to their electron microscopefacilities, and Ralf Lendt (University of Hamburg) for compound-specific carbon isotope measurements. We thank Bram Vekeman forproviding the genomes from the MMG2 North Sea enrichments. TheCampeche Knoll cruise was funded by the German Research Foun-dation (DFG – Deutsche Forschungsgemeinschaft). Additional supportwas provided through the MARUM DFG-Research Center/ExcellenceCluster “The Ocean in the Earth System” at the University of Bremen.We are grateful to the Mexican authorities for granting permission toconduct this research in the southern Gulf of Mexico (permission ofDGOPA: 02540/14 from 5 November 2014). This study was fundedby the Max Planck Society, the MARUM DFG-Research Center/Excellence Cluster “The Ocean in the Earth System” at the Universityof Bremen, an ERC Advanced Grant (BathyBiome, 340535) and a
Table 2 Relative composition oflipid biomarkers and theiraverage δ13C values in spongetissue
lipid biomarker sources H. (S.) methanophila I. methanophila
% of all lipids δ13C (‰) (av.) % of all lipids δ13C (‰) (av.)
MOX (MUFAs) 27 −46 14 −50
Sponge (demospongic acids, MUFAs) 28 −47 39 −43
Sponge (sterols) 29 −41 26 −43
Various sources (saturated n-fatty acids) 12 −36 16 −35
Bacteria, usually SRB (tb fatty acids) 1 NM 1 NM
Various bacteria (diplopterol) 3 −47 4 −43
MOX methane-oxidizing bacteria, SRB sulfate-reducing bacteria, tb terminally-branched, MUFAsmonounsaturated fatty acids, av. average, NM not measured
1222 M. Rubin-Blum et al.
Gordon and Betty Moore Foundation Marine Microbial InitiativeInvestigator Award to ND (Grant GBMF3811), the DFG CollaborativeResearch Center 1182 ‘Origin and Function of Metaorganisms to UHand ND, and the European Union’s Horizon 2020 research andinnovation program to PC and UH under Grant Agreement No.679849 (‘SponGES’). CPA was supported by a postdoctoral fellow-ship from the Alexander von Humboldt Foundation.
Author contributions MRB, CPA, DB, JP, DB, HS, UH, and NDconceived the study. HS provided the framework for deep-sea samplecollections. YM and IMD provided in situ documentation. MRB,CPA, LS, YCW, CMP, DB, JP, and PC analyzed the samples. MRBwrote the manuscript with contributions from all co-authors.
Compliance with ethical standards
Conflict of interest The authors declare that they have no conflict ofinterest.
Publisher’s note: Springer Nature remains neutral with regard tojurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative CommonsAttribution 4.0 International License, which permits use, sharing,adaptation, distribution and reproduction in any medium or format, aslong as you give appropriate credit to the original author(s) and thesource, provide a link to the Creative Commons license, and indicate ifchanges were made. The images or other third party material in thisarticle are included in the article’s Creative Commons license, unlessindicated otherwise in a credit line to the material. If material is notincluded in the article’s Creative Commons license and your intendeduse is not permitted by statutory regulation or exceeds the permitteduse, you will need to obtain permission directly from the copyrightholder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.
References
1. Webster NS. Cooperation, communication, and co-evolution:grand challenges in microbial symbiosis research. Front Micro-biol. 2014;5:164.
2. Schmitt S, Tsai P, Bell J, Fromont J, Ilan M, Lindquist N, et al.Assessing the complex sponge microbiota: core, variable andspecies-specific bacterial communities in marine sponges. ISME J.2012;6:564–76.
3. Thomas T, Moitinho-Silva L, Lurgi M, Björk JR, Easson C,Astudillo-García C, et al. Diversity, structure and convergentevolution of the global sponge microbiome. Nat Commun.2016;7:11870.
4. Taylor MW, Radax R, Steger D, Wagner M. Sponge-associatedmicroorganisms: evolution, ecology, and biotechnological poten-tial. Microbiol Mol Biol Rev. 2007;71:295–347.
5. Gloeckner V, Wehrl M, Moitinho-silva L, Schupp P, Pawlik JR,Lindquist NL, et al. The HMA-LMA dichotomy revisited: anelectron microscopical survey of 56 sponge species. Biol Bull.2017;227:78–88.
6. Webster NS, Thomas T. The sponge hologenome. MBio. 2016;7:e00135-16.
7. Hentschel U, Piel J, Degnan SM, Taylor MW. Genomic insightsinto the marine sponge microbiome. Nat Rev Microbiol.2012;10:641–54.
8. Wilson MC, Mori T, Rückert C, Uria AR, Helf MJ, Takada K,et al. An environmental bacterial taxon with a large and distinctmetabolic repertoire. Nature. 2014;506:58–62.
9. Burgsdorf I, Slaby BM, Handley KM, Haber M, Blom J, MarshallCW, et al. Lifestyle evolution in cyanobacterial symbionts ofsponges. mBio. 2015;6:e00391–15.
10. Gauthier M-EA, Watson JR, Degnan SM. Draft genomes shedlight on the dual bacterial symbiosis that dominates the micro-biome of the coral reef sponge Amphimedon queenslandica. FrontMar Sci. 2016;3:00196.
11. Slaby BM, Hackl T, Horn H, Bayer K, Hentschel U. Metagenomicbinning of a marine sponge microbiome reveals unity in defensebut metabolic specialization. ISME J. 2017;11:2465–78.
12. Ryu T, Seridi L, Moitinho-Silva L, Oates M, Liew YJ, Mavro-matis C, et al. Hologenome analysis of two marine sponges withdifferent microbiomes. BMC Genom. 2016;17:158.
13. Fan L, Reynolds D, Liu M, Stark M, Kjelleberg S, Webster NS,et al. Functional equivalence and evolutionary convergence incomplex communities of microbial sponge symbionts. Proc NatlAcad Sci USA. 2012;109:E1878–E1887.
14. Gao Z-M, Wang Y, Tian R-M, Wong YH, Batang ZB, Al-Suwailem AM, et al. Symbiotic adaptation drives genomestreamlining of the cyanobacterial sponge symbiont “CandidatusSynechococcus spongiarum”. mBio. 2014;5:e00079–14.
15. Lackner G, Peters EE, Helfrich EJN, Piel J. Insights into thelifestyle of uncultured bacterial natural product factories asso-ciated with marine sponges. Proc Natl Acad Sci USA. 2017;114:E347–E356.
16. Liu M, Fan L, Zhong L, Kjelleberg S, Thomas T. Metaproteo-genomic analysis of a community of sponge symbionts. ISME J.2012;6:1515–25.
17. Moitinho-Silva L, Díez-Vives C, Batani G, Esteves AI, Jahn MT,Thomas T. Integrated metabolism in sponge–microxbe symbiosisrevealed by genome-centered metatranscriptomics. ISME J.2017;11:1651–66.
18. Fiore CL, Labrie M, Jarett JK, Lesser MP. Transcriptional activityof the giant barrel sponge, Xestospongia muta holobiont: mole-cular evidence for metabolic interchange. Front Microbiol. 2015;6:364.
19. Radax R, Rattei T, Lanzen A, Bayer C, Rapp HT, Urich T, et al.Metatranscriptomics of the marine sponge Geodia barretti: tack-ling phylogeny and function of its microbial community. EnvironMicrobiol. 2012;14:1308–24.
20. Moitinho-Silva L, Seridi L, Ryu T, Voolstra CR, Ravasi T,Hentschel U. Revealing microbial functional activities in the RedSea sponge Stylissa carteri by metatranscriptomics. EnvironMicrobiol. 2014;16:3683–98.
21. Mori T, Cahn JKB, Wilson MC, Meoded RA, Wiebach V,Martinez AFC. et al. Single-bacterial genomics validatesrich and varied specialized metabolism of uncultivatedEntotheonella sponge symbionts. Proc Natl Acad Sci USA.2018;115:1718–23.
22. Jørgensen BB, Boetius A. Feast and famine—microbial life in thedeep-sea bed. Nat Rev Microbiol. 2007;5:770–81.
23. Vacelet J, Duport E. Prey capture and digestion in the carnivoroussponge Asbestopluma hypogea (Porifera: Demospongiae). Zoo-morphology. 2004;123:179–90.
24. Verhoeven JTP, Kavanagh AN, Dufour SC. Microbiome analysisshows enrichment for specific bacteria in separate anatomicalregions of the deep-sea carnivorous sponge Chondrocladiagrandis. FEMS Microbiol Ecol. 2017;93:fiw214.
25. Hestetun JT, Dahle H, Jørgensen SL, Olsen BR, Rapp HT. TheMicrobiome and occurrence of methanotrophy in carnivoroussponges. Front Microbiol. 2016;7:1781.
26. Arellano SM, Lee OO, Lafi FF, Yang J, Wang Y, Young CM,et al. Deep sequencing of Myxilla (Ectyomyxilla) methanophila,an epibiotic sponge on cold-seep tubeworms, reveals methylo-trophic, thiotrophic, and putative hydrocarbon-degrading micro-bial associations. Microb Ecol. 2013;65:450–61.
Fueled by methane: deep-sea sponges from asphalt seeps gain their nutrition from methane-oxidizing. . . 1223
27. Tian R-M, Zhang W, Cai L, Wong Y-H, Ding W, Qian P-Y, et al.Genome reduction and microbe-host interactions drive adaptationof a sulfur-oxidizing bacterium associated with a cold seepsponge. mSystems. 2017;2:e00184–16.
28. Nishijima M, Lindsay DJ, Hata J, Nakamura A, Kasai H, Ise Y,et al. Association of thioautotrophic bacteria with deep-seasponges. Mar Biotechnol. 2010;12:253–60.
29. Petersen JM, Wentrup C, Verna C, Knittel K, Dubilier N. Originsand evolutionary flexibility of chemosynthetic symbionts fromdeep-sea animals. Biol Bull. 2012;223:123–37.
30. Vacelet J, Boury-Esnault N, Fiala-Medioni A, Fisher CR. Amethanotrophic carnivorous sponge. Nature. 1995;377:296.
31. Dubilier N, Bergin C, Lott C. Symbiotic diversity in marine ani-mals: the art of harnessing chemosynthesis. Nat Rev Microbiol.2008;6:725–40.
32. Sahling H, Borowski C, Escobar-Briones E, Gaytán-Caballero A,Hsu C-W, Loher M, et al. Massive asphalt deposits, oil seepage,and gas venting support abundant chemosynthetic communities atthe Campeche Knolls, southern Gulf of Mexico. Biogeosciences.2016;13:4491–512.
33. Rubin-Blum M, Antony CP, Borowski C, Sayavedra L, Pape T,Sahling H, et al. Short-chain alkanes fuel mussel and spongeCycloclasticus symbionts from deep-sea gas and oil seeps. NatMicrobiol. 2017;17093:17093.
34. Montanaro J, Gruber D, Leisch N. Improved ultrastructure ofmarine invertebrates using non-toxic buffers. PeerJ. 2016;4:e1860.
35. Kumar Y, Westram R, Kipfer P, Meier H, Ludwig W. Evaluationof sequence alignments and oligonucleotide probes with respect tothree-dimensional structure of ribosomal RNA using ARB soft-ware package. BMC Bioinforma. 2006;7:240.
36. Ruff SE, Arnds J, Knittel K, Amann R, Wegener G, Ramette A,et al. Microbial communities of deep-sea methane seeps atHikurangi continental margin (New Zealand). PLoS ONE. 2013;8:e72627.
37. Duperron S, Halary S, Lorion J, Sibuet M, Gaill F. Unexpectedco-occurrence of six bacterial symbionts in the gills of the coldseep mussel Idas sp. (Bivalvia: Mytilidae). Environ Microbiol.2008;10:433–45.
38. LŞ Yılmaz, Noguera DR. Development of thermodynamic modelsfor simulating probe dissociation profiles in fluorescence in situhybridization. Biotechnol Bioeng. 2007;96:349–63.
39. Amann RI, Blinder BJ, Olson RJ, Chisholm SW, Devereux R,Stahl DA. Combination of 16S rRNA-targeted oligonucleotideprobes with flow cytometry for analyzing mixed microbialpopulations. Appl Environ Microbiol. 1990;56:1919–25.
40. Wallner G, Amann R, Beisker W. Optimizing fluorescent in situhybridization with rRNA-targeted oligonucleotide probes for flowcytometric identification of microorganisms. Cytometry.1993;14:136–43.
41. Peng Y, Leung HCM, Yiu SM, Chin FYL. IDBA-UD: a de novoassembler for single-cell and metagenomic sequencing data withhighly uneven depth. Bioinformatics. 2012;28:1420–8.
42. Seah BKB, Gruber-Vodicka HR. gbtools: interactive visualizationof metagenome bins in R. Front Microbiol. 2015;6:1451.
43. Nurk S, Bankevich A, Antipov D, Gurevich A, Korobeynikov A,Lapidus A, et al. Assembling genomes and mini-metagenomes fromhighly chimeric reads. In: Deng M, Jiang R, Sun F, Zhang X,editors. Res Comput Mol Biol. RECOMB, Lecture Notes in Com-puter Science, Springer; Berlin, Heidelberg. 2013;7821:158–70.
44. Bankevich A, Nurk S, Antipov D, Gurevich AA, Dvorkin M,Kulikov AS, et al. SPAdes: a new genome assembly algorithmand its applications to single-cell sequencing. J Comput Biol.2012;19:455–77.
45. Gurevich A, Saveliev V, Vyahhi N, Tesler G. QUAST: qualityassessment tool for genome assemblies. Bioinformatics.2013;29:1072–5.
46. Parks DH, Imelfort M, Skennerton CT, Hugenholtz P, Tyson GW.CheckM: assessing the quality of microbial genomes recoveredfrom isolates, single cells, and metagenomes. Genome Res.2015;25:1043–55.
47. Liao Y, Smyth GK, Shi W. FeatureCounts: an efficient generalpurpose program for assigning sequence reads to genomic fea-tures. Bioinformatics. 2014;30:923–30.
48. Wagner GP, Kin K, Lynch VJ. Measurement of mRNA abun-dance using RNA-seq data: RPKM measure is inconsistent amongsamples. Theory Biosci. 2012;131:281–5.
49. Grabherr MG, Haas BJ, Yassour M, Levin JZ, Thompson DA,Amit I, et al. Full-length transcriptome assembly from RNA-Seqdata without a reference genome. Nat Biotechnol. 2011;29:644–52.
50. Li B, Dewey CN. RSEM: accurate transcript quantification fromRNA-Seq data with or without a reference genome. BMCBioinforma. 2011;12:323.
51. Wu M, Scott AJ. Phylogenomic analysis of bacterial and archaealsequences with AMPHORA2. Bioinformatics. 2012;28:1033–4.
52. Edgar RC. MUSCLE: multiple sequence alignment with highaccuracy and high throughput. Nucleic Acids Res. 2004;32:1792–7.
53. Kumar S, Stecher G, Tamura K. MEGA7: molecular evolutionarygenetics analysis version 7.0 for bigger datasets. Mol Biol Evol.2016;33:msw054.
54. Le SQ, Gascuel O. An improved general amino acid replacementmatrix. Mol Biol Evol. 2008;25:1307–20.
55. Nichols PD, Guckert JB, White DC. Determination of mono-saturated fatty acid double-bond position and geometry formicrobial monocultures and complex consortia by capillary GC-MS of their dimethyl disulphide adducts. J Microbiol Methods.1986;5:49–55.
56. Brüning M, Sahling H, MacDonald IR, Ding F, Bohrmann G.Origin, distribution, and alteration of asphalts at Chapopote Knoll,Southern Gulf of Mexico. Mar Pet Geol. 2010;27:1093–106.
57. Moitinho-Silva L, Steinert G, Nielsen S, Hardoim CCP, Wu YC,McCormack GP, et al. Predicting the HMA-LMA status in marinesponges by machine learning. Front Microbiol. 2017;8:1–14.
58. Vacelet J, Donadey C. Electron microscope study of the asso-ciation between some sponges and bacteria. J Exp Mar Bio Ecol.1977;30:301–14.
59. Maldonado M. Intergenerational transmission of symbiotic bac-teria in oviparous and viviparous demosponges, with emphasis onintracytoplasmically-compartmented bacterial types. J Mar BiolAssoc U Kingd. 2007;87:1701–13.
60. Jain C, Rodriguez-R LM, Phillippy AM, Konstantinidis KT,Aluru S. High-throughput ANI analysis of 90K prokaryotic gen-omes reveals clear species boundaries. Nat Commun.2018;9:5114.
61. Parks DH, Chuvochina M, Waite DW, Rinke C, Skarshewski A,Chaumeil P-A, et al. A standardized bacterial taxonomy based ongenome phylogeny substantially revises the tree of life. Nat Bio-technol. 2018;36:996–1004.
62. Vekeman B, Speth D, Wille J, Cremers G, De Vos P, Op denCamp HJM, et al. Genome characteristics of two novel type Imethanotrophs enriched from North Sea sediments containingexclusively a lanthanide-dependent XoxF5-type methanol dehy-drogenase. Microb Ecol. 2016;72:503–9.
63. Pasulka AL, Goffredi SK, Tavormina PL, Dawson KS, Levin LA,Rouse GW, et al. Colonial tube-dwelling ciliates influencemethane cycling and microbial diversity within methane seepecosystems. Front Mar Sci. 2017;3:276.
64. Watsuji T, Nakagawa S, Tsuchida S, Toki T, Hirota A, TsunogaiU, et al. Diversity and function of epibiotic microbial communitieson the galatheid crab. Shinkaia crosnieri Microbes Environ.2010;25:288–94.
1224 M. Rubin-Blum et al.
65. Rodrigues CF, Hilário A, Cunha MR, Weightman AJ, Webster G.Microbial diversity in Frenulata (Siboglinidae, Polychaeta) speciesfrom mud volcanoes in the Gulf of Cadiz (NE Atlantic). AntonieVan Leeuwenhoek, Int J Gen Mol Microbiol. 2011;100:83–98.
66. Petersen JM, Dubilier N. Methanotrophic symbioses in marineinvertebrates. Environ Microbiol Rep. 2009;1:319–35.
67. Vacelet J, Fiala-Médioni A, Fisher CR, Boury-Esnault N. Sym-biosis between methane-oxidizing bacteria and a deep-sea carni-vorous cladorhizid sponge. Mar Ecol Prog Ser. 1996;145:77–85.
68. Padilla CC, Bertagnolli AD, Bristow LA, Sarode N, Glass JB,Thamdrup B, et al. Metagenomic binning recovers a tran-scriptionally active gammaproteobacterium linking methano-trophy to partial denitrification in an anoxic oxygen minimumzone. Front Mar Sci. 2017;4:23.
69. Moran NA, Plague GR. Genomic changes following hostrestriction in bacteria. Curr Opin Genet Dev. 2004;14:627–33.
70. Moran NA, Bennett GM. The tiniest tiny genomes. Annu RevMicrobiol. 2014;68:195–215.
71. Salem H, Florez L, Gerardo N, Kaltenpoth M. An out-of-bodyexperience: the extracellular dimension for the transmission ofmutualistic bacteria in insects. Proc R Soc B Biol Sci.2015;282:20142957–20142957.
72. Wernegreen JJ. Endosymbiont evolution: predictions from theoryand surprises from genomes. Ann N Y Acad Sci. 2015;1360:16–35.
73. Batut B, Knibbe C, Marais G, Daubin V. Reductive genomeevolution at both ends of the bacterial population size spectrum.Nat Rev Microbiol. 2014;12:841–50.
74. Mann S, Chen YPP. Bacterial genomic G+C composition-eliciting environmental adaptation. Genomics. 2010;95:7–15.
75. Anthony C. The biochemistry of methylotrophs. London, NewYork: Academic Press; 1982.
76. Kalyuzhnaya MG, Yang S, Rozova ON, Smalley NE, Clubb J,Lamb A, et al. Highly efficient methane biocatalysis revealed in amethanotrophic bacterium. Nat Commun. 2013;4:2785.
77. Ponnudurai R, Kleiner M, Sayavedra L, Petersen JM, Moche M,Otto A, et al. Metabolic and physiological interdependencies inthe Bathymodiolus azoricus symbiosis. ISME J. 2017;11:463–77.
78. Hendry TA, Freed LL, Fader D, Fenolio D, Sutton TT, Lopez JV.Ongoing transposon-mediated genome reduction in the luminousbacterial symbionts of deep-sea ceratioid anglerfishes. mBio.2018;9:e01033–18.
79. Oakeson KF, Gil R, Clayton AL, Dunn DM, Von NiederhausernAC, Hamil C, et al. Genome degeneration and adaptation in anascent stage of symbiosis. Genome Biol Evol. 2014;6:76–93.
80. Kleiner M, Young JC, Shah M, Verberkmoes NC, Dubilier N.Metaproteomics reveals abundant transposase expression inmutualistic endosymbionts. mBio. 2013;4:16–18.
81. Kleiner M, Wentrup C, Lott C, Teeling H, Wetzel S, Young J,et al. Metaproteomics of a gutless marine worm and its symbioticmicrobial community reveal unusual pathways for carbon andenergy use. Proc Natl Acad Sci USA. 2012;109:1173–82.
82. Van Melderen L. Toxin-antitoxin systems: Why so many, whatfor? Curr Opin Microbiol. 2010;13:781–5.
83. Cimermancic P, Medema MH, Claesen J, Kurita K, WielandBrown LC, Mavrommatis K, et al. Insights into secondary meta-bolism from a global analysis of prokaryotic biosynthetic geneclusters. Cell. 2014;158:412–21.
84. Díez-Vives C, Moitinho-Silva L, Nielsen S, Reynolds D, ThomasT. Expression of eukaryotic-like protein in the microbiome ofsponges. Mol Ecol. 2017;26:1432–51.
85. Desvaux M, Parham NJ, Henderson IR. The autotransportersecretion system. Res Microbiol. 2004;155:53–60.
86. Kellermann MY, Schubotz F, Elvert M, Lipp JS, Birgel D, Prieto-Mollar X, et al. Symbiont–host relationships in chemosyntheticmussels: a comprehensive lipid biomarker study. Org Geochem.2012;43:112–24.
87. Summons RE, Jahnke LL, Roksandic Z. Carbon isotopic frac-tionation in lipids from methanotrophic bacteria: relevance forinterpretation of the geochemical record of biomarkers. GeochimCosmochim Acta. 1994;58:2853–63.
88. Jahnke LL, Summons RE, Hope JM, Des Marais DJ. Carbonisotopic fractionation in lipids from methanotrophic bacteria II:the effects of physiology and environmental parameters on thebiosynthesis and isotopic signatures of biomarkers. GeochimCosmochim Acta. 1999;63:79–93.
Fueled by methane: deep-sea sponges from asphalt seeps gain their nutrition from methane-oxidizing. . . 1225